Comparison of three different networks at level 1 using reported protein-protein interactions vs. using inferred interactions by sequence homology. A BIANA database has been created using the following databases: Uniprot Swissprot, IntAct, MINT, BioGrid, DIP and HPRD. Three different initial data sets related with three different pathologies have been created by a keyword search in fields Disease, Keyword, Description and Function. Two networks at level 1 have been created for each set: 1) using reported protein-protein interactions by third-party databases and 2) using inferred interactions by using sequence similarity (see text for details). For each network we calculated the number of proteins involved in the pathologies according to HEFalMp  with a p < 0.00001 (shown in parenthesis). By using inferred interactions a higher number of candidates are retrieved.